spatial-cross-validation

The blockCV package creates spatially or environmentally separated training and testing folds for cross-validation to provide a robust error estimation in spatially structured environments. See

Spatial cross-validation for GeoAI

Draw map boundaries your data supports instead of inheriting ones that don't fit. spatialkit tessellates point observations into Voronoi, hex, grid or Delaunay cells, aggregates to them with autocorrelation-aware standard errors, and fits GWR, Bayesian GP or random-forest models validated by spatial cross-validation.

soilcnn — convolutional neural networks for digital soil mapping in R: spatial cross-validation, seed ensembles, conformal intervals and area of applicability, from rasters to maps.

Predicting which Philippine municipalities stay without power after typhoons, using NASA Black Marble nighttime lights, Google Earth Engine and machine learning, with spatially explicit tests of transfer to unseen storms.

Spatially validated machine-learning workflow for mineral prospectivity mapping using geochemistry, structural geology, magnetics and gravity.

BathySurrogate: An Open-Source Environmental Surrogate Framework for Satellite-Derived Bathymetry via Multi-Source Data Fusion and Spatial Validation

Spatially explicit, interpretable modelling of land surface temperature across Awka, Onitsha and Nnewi, Anambra State, Nigeria. Landsat 9 predictors in Google Earth Engine, variogram-derived blocked cross-validation, and SHAP attribution of urban heat drivers.

Data and code for pixel-based landslide susceptibility modelling in Huancabamba (Peru) using Random Forest and spatial block cross-validation.

CLIMORFA: reproducible analysis pipeline for climate-sensitive urban morphology and Local Climate Zone classification in Izmir

Reproducibility package for 'Physics-Consistent Multi-Target RF Field Reconstruction over Heterogeneous Terrain Graphs under Spatially Blocked Evaluation' (IEEE Access): code, results, and seed-42 checkpoints.

Spatially explicit modelling of post-fire vegetation recovery across 15 California wildfires (2018–2022) using Sentinel-2, terrain, and climate.

Predict rarely measured soil properties (total N, CEC7) by learning the bounded parameters of known soil relations; spatial benchmark with QRF, TabM and TabICLv2 on public WoSIS data.

MSc dissertation workflow for predicting Leeds neighbourhood deprivation from Google Street View imagery using spatial validation and explainable machine learning.

Classificação de uso e ocupação do solo (LULC) em Vieira do Minho, Portugal: Sentinel-1/2, LiDAR e Random Forest, com validação cruzada espacial e auditoria metodológica documentada. Portfólio técnico.

Machine learning framework for lithium concentration prediction and spatial exploration prioritization using geochemical data, Sentinel-2 imagery, spatial validation, uncertainty checks, and reproducible training pipelines.

Benchmark for cross-survey evaluation of geochemical machine learning under detection-limit censoring: four harmonised surveys, censoring metadata, frozen validation folds and evaluation harness.

Spatial cross-validated Random Forest software for reproducible landslide susceptibility modelling and mapping.

Reading Berlin's rent-index residential tiers from aerial imagery — and why the spatial split is half the result. Deep Learning course project, 2026.

Türkiye Çok Ölçekli Taşkın Riski İndeksi Sınıflandırma ve Mekansal Genellenebilirlik Platformu (Water 2026, 18(20), 2515 uyarlaması)